R MICE Impute missing data conditionally on group allocation

Question

I am using mice in R to impute missing data for an intervention study. The data I want to impute are continuous. I have two observation groups; 1= the intervention group, 2= the control group. There are some variables that I only want mice to impute for the intervention group. For the control group, the NA's should be remaining.

Here is a toy version of my data set:

toy2 <- data.frame(group= c(1, 1, 2, 1, 2, 1, 2, 1, 2, 2, 1, 1, 2, 1, 2),
                  PSS_t1=c(16, 15, 17, 22, 30, 12, 18, 19, 20, 14, 21, 23, 27, 20, 14),
                  CSQ_t2= c(8, 9, NA, 12, NA, 7, NA, 9, NA, NA, 10, 11, NA, 6, NA),
                  PSS_t3=c(15, 14, 15, 19, 22, 7, 19, 21, 14, 18, 12, 10, 22, 17, 14),
                  PSS_t4=c(1, 1, NA, 1, NA, 1, NA, 1, NA, NA, 1, 1, NA, 1, NA))

The scales CSQ_t2 and PSS_t4 have only been presented to the intervention group, so NA's for the control group are natural.

I thought about using post from mice, unfortunately it does not work:

posttoy2 <- mice(toy2, m = 1, maxit = 0)$post
posttoy2["CSQ_t2"] <- "ifdo(group==1)"
posttoy2["PSS_t4"] <- "ifdo(group==1)"

imptoy2 <- mice(toy2, m = 1, post = posttoy2, seed = 555,
                      print = TRUE)

Here is the output I get:

> toy2 <- data.frame(group= c(1, 1, 2, 1, 2, 1, 2, 1, 2, 2, 1, 1, 2, 1, 2),
+                   PSS_t1=c(16, 15, 17, 22, 30, 12, 18, 19, 20, 14, 21, 23, 27, 20, 14),
+                   CSQ_t2= c(8, 9, NA, 12, NA, 7, NA, 9, NA, NA, 10, 11, NA, 6, NA),
+                   PSS_t3=c(15, 14, 15, 19, 22, 7, 19, 21, 14, 18, 12, 10, 22, 17, 14),
+                   PSS_t4=c(1, 1, NA, 1, NA, 1, NA, 1, NA, NA, 1, 1, NA, 1, NA))
> posttoy2 <- mice(toy2, m = 1, maxit = 0)$post
Warning message:
Number of logged events: 1 
> posttoy2["CSQ_t2"] <- "ifdo(group_unguided==1)"
> posttoy2["CSQ_t2"] <- "ifdo(group==1)"
> posttoy2["PSS_t4"] <- "ifdo(group==1)"
> posttoy2
           group           PSS_t1           CSQ_t2           PSS_t3           PSS_t4 
              ""               "" "ifdo(group==1)"               "" "ifdo(group==1)" 
> imptoy2 <- mice(toy2, m = 1, post = posttoy2, seed = 555,
+                       print = TRUE)                        

 iter imp variable
  1   1  CSQ_t2Function ifdo() not yet implemented.

  2   1  CSQ_t2Function ifdo() not yet implemented.

  3   1  CSQ_t2Function ifdo() not yet implemented.

  4   1  CSQ_t2Function ifdo() not yet implemented.

  5   1  CSQ_t2Function ifdo() not yet implemented.

Warning message:
Number of logged events: 6

Has someone any idea how I can realize this conditional imputation?

Many thanks in advance for your ideas and help! :-)

R MICE Impute missing data conditionally on group allocation

0 Answers